The present invention relates to an exhaust system for a lean-burn internal combustion engine comprising a NOx-absorbent and, in particular, to a method of controlling reductant addition into the exhaust system for the purpose of regenerating the NOx-absorbent and reducing NOx to N2.

An exhaust system for a lean-burn internal combustion engine such as a diesel engine or a lean-burn gasoline engine comprising a NOx-absorbent is known from, for example, EP 0560991.

As used herein, a "NOx-trap" is a catalyst comprising a NOx-absorbent and a catalyst for oxidising NO to NO2. NOx-traps are also known as "lean NOx traps" or "LNC".

A typical NOx-trap formulation includes a catalytic oxidation component, such as Pt, a NOx-absorbent, such as compounds of alkali metals e.g. potassium and/or caesium; compounds of alkaline earth metals, such as barium or strontium; or compounds of rare- earth metals, typically lanthanum and/or yttrium; and a reduction catalyst, e.g. rhodium. One mechanism commonly given for NOx-storage during lean engine operation for this formulation is that, in a first step, the NO reacts with oxygen on active oxidation sites on the Pt to form NO2. The second step involves adsorption of the NO2 by the storage material in the form of an inorganic nitrate.

Whilst the inorganic NOx-storage component is typically present as an oxide, it is understood that in the presence of air or exhaust gas containing CO2 and H2O it may also be in the form of the carbonate or possibly the hydroxide.

When the engine runs intermittently under enriched conditions or at elevated temperatures, the nitrate species become thermodynamically unstable and decompose, producing NO or NO2. Under richer conditions, these NOx species are reduced by carbon monoxide, hydrogen and hydrocarbons to N2, which can take place over the reduction catalyst.

An object of an exhaust system comprising a NOx-trap is to improve the economy of the engine whilst meeting the relevant emissions standard, e.g. Euro IV.

Systems to control reductant addition for the purpose of regenerating a NOx-trap and reducing desorbed NOx are known, but tend to require very complicated control regimes involving multiple sensor inputs and processors to run complex algorithms. As a result, such systems are very expensive.

EP-B-0341832 (incorporated herein by reference) describes a process for combusting particulate matter (PM) in diesel exhaust gas, which method comprising oxidising NO in the exhaust gas to NO2 on a catalyst, filtering the PM from the exhaust gas and combusting the filtered PM in the NO2 at up to 400°C. Such a system is available from Johnson Matthey and is marketed as the CRT®.

We have investigated methods of regenerating NOx-absorbents and we have discovered that it is possible to meet a relevant emission standard, such as Euro FV, with an exhaust system comprising a NOx-absorbent without the need for complex equipment such as algorithm-programmed processors and a network of sensor inputs. Such a discovery has particular application to the retrofit market.

According to a first aspect of the invention, there is provided an exhaust system for a vehicular lean-burn internal combustion engine comprising a NOx-absorbent, a reductant injector disposed upstream of the NOx-absorbent and means, when in use, for controlling reductant addition, wherein the reductant addition control means supplies reductant to the NOx-absorbent at all vehicle speeds in a duty cycle at a rate which is predetermined to correlate with a desired NOx conversion at the average duty cycle speed of the vehicle.

The invention of the first aspect has particular application to the retrofit market for vehicles of a limited duty cycle such as buses or refuse trucks. The idea is to determine what rate of reductant injection is required to reduce a chosen quantity OfNOx, e.g. 90%, in a NOx-absorbent at the average duty cycle speed. For example, when the NOx-absorbent is a component of a NOx-trap, the system controller can be arranged, when in use, to generate a continuous tempo and quantity of hydrocarbon (HC) fuel injection e.g. injection at 2 seconds every minute. The system controller can also be arranged to provide occasional relatively long rich HC fuel pulses to ensure that the NOχ-trap is substantially completely regenerated, followed by the more frequent sequence of shorter enrichment pulses to maintain the storing capability of the NOx-trap. The exact detail of the injection strategy depends on the vehicle and its duty cycle.

At speeds higher than the average duty cycle speed, there would be more NOx and a greater mass airflow and so NOx conversion overall would fall off, because there would be insufficient reductant. However, because higher speed would be less likely e.g. in city centre buses, the system can meet NOx emission standards over an entire drive cycle without increasing fuel penalty; equally where the vehicle speed drops below the average duty cycle speed, HC can be emitted, but on average over a duty cycle the system can meet the emission standard for HC. The correlation of the rate of HC injection to average duty cycle speed can be tailored to the particular application, e.g. buses in Manchester (UK) city centre would be expected to encounter different duty cycles to those in London (UK) city centre.

In one embodiment of the first aspect, an oxidation catalyst is disposed between the reductant injector and the NOx-absorbent for increasing the temperature of the NOx-trap for regeneration and/or to remove oxygen from the exhaust gas to ensure a rich exhaust gas for regeneration of the NOx-absorbent.

In a particular arrangement, the NOx-trap and systems for delivering reductant described herein are disposed downstream of the arrangement described in EP-B-0341832, mentioned hereinabove. That is, a catalyst for oxidising NO to NO2 is followed by an optionally catalysed filter then a reductant injector followed by the NOx-absorbent.

In one embodiment, the NOx-absorbent for use in the invention is a component of a NOx-trap. Unless otherwise described, the catalysts for use in the present invention are coated on high surface area substrate monoliths made from metal or ceramic or silicon carbide, e.g. cordierite, materials. A common arrangement is a honeycomb, flow- through monolith structure of from 100-600 cells per square inch (cpsi) such as 300-400 cpsi (15.5-93.0 cells cm"2, e.g. 46.5-62.0 cells cm"2).

The internal combustion engine can be a diesel or lean-burn gasoline engine, such as a gasoline direct injection engine. The diesel engine can be a light-duty engine or a heavy-duty engine, as defined by the relevant legislation.

A method of reducing NOx in the exhaust gas of a vehicular lean-burn internal combustion engine according to a second aspect of the invention comprises absorbing NOx from the exhaust gas in a NOx absorbent, contacting the NOx absorbent with a reductant to regenerate the NOx-absorbent at all vehicle speeds in a duty cycle, and reducing NOx to N2, wherein a rate of reductant injection correlates with a desired NOx conversion at the average duty cycle speed.

In order that the present invention may be more fully understood, embodiments thereof will now be described with reference to the accompanying drawings, in which:

Figure 1 shows a schematic system according to the first aspect of the invention;

Figure 2 is a schematic graph plotting quantity of fuel against time showing a fuel injection strategy for use in the system of Figure 1 ;

Figure 3 is a schematic of a working embodiment of the invention;

Figure 4 is a graph showing the upstream Air/Fuel Ratio (AFR) as a function of road speed in the embodiment of Figure 3;

Figure 5 is a graph showing NOx measurements at the idle condition for the embodiment of Figure 3; Figure 6 is a graph showing the corresponding system temperatures at the idle condition for the trace shown in Figure 5;

Figure 7 is a graph showing NOx measurements at 40mph for the embodiment of Figure 3;

Figure 8 is a graph showing the corresponding temperature measurements at 40mph for the trace shown in Figure 7; and

Figure 9 is a graph showing the NOx conversion as a function of road speed for the system of Figure 3.

In the system 50 depicted in Figure 1, 52 is a conditional system controller (CSC), 54 is a master switch, 56 is an alternator, 58 is a blocking capacitor, 60 is a thermocouple, 62 is an injection controller (ICU), 64 is a fuel pump, 66 is a valve, 68 is a fuel injector and 70 is a positive power line. The CSC 52 is a switch providing power to the ICU 62 if the master power switch 54 is on, the engine is running as determined by an AC ripple from the alternator 56 present after a DC blocking capacitor 58 and the output of a suitably placed thermocouple 60 to detect the exhaust system is above a minimum pre-determined temperature for reduction of NOx on a suitable NOx-trap. The master switch 54 need not be connected to the key-on position.

The CSC 52 is designed to generate a continuous tempo and quantity of HC injection when all three features (master switch position, detection of alternator ripple and exhaust gas temperature above a pre-determined minimum) coincide. When the CSC 52 is on, power is supplied to the injection pump 64 and the ICU 62 that operates a solenoid valve 66 to produce a series of pulses to enrich the exhaust gas before it passes over an oxidation catalyst upstream of the NOx absorbing components. Typically the injection controller will provide occasional relatively very long rich pulses to ensure that the NOx-trap is substantially completely empty and this is followed by a more frequent sequence of shorter enrichment pulses, e.g. injection at 2 seconds every minute, to maintain the storing capability of the NOx-trap (see Figure 2). This fuel injection rate is correlated to a chosen NOx conversion e.g. 90% at the average duty cycle speed. The exact detail of the injection strategy depends on the vehicle and its duty cycle.

Whilst, very generally, the systems employing NOx-traps described herein have been developed to provide simple control mechanisms to predict when NOx-trap regeneration should be done, with particular application to retrofit, many vehicles already include a range of sensors to input data to the ECU for controlling other aspects of vehicular operation. By suitable re-programming of the ECU it is possible to adopt one or more of such existing sensor inputs for the purposes of predicting remaining NOx-trap capacity. These include, but are not limited to, predetermined or predicted time elapsed from key-on or previous regeneration, by sensing the status of a suitable clock means; airflow over the TWC or manifold vacuum; ignition timing; engine speed; throttle position; exhaust gas redox composition, for example using a lambda sensor, preferably a linear lambda sensor; quantity of fuel injected in the engine; where the vehicle includes an exhaust gas recirculation (EGR) circuit, the position of the EGR valve and thereby the detected amount of EGR; engine coolant temperature; and where the exhaust system includes a NOx sensor, the amount of NOx detected upstream and/or downstream of the NOx-trap. Where the clock embodiment is used, the predicted time can be subsequently adjusted in response to data input.

The following specific Example is provided by way of illustration only.

Example

The exhaust system (10) (shown in Figure 3) of a single deck bus fitted with a 6 litre turbocharged engine and comprising engine turbo (12), type approved to European Stage 1 emission limits, was modified to incorporate a three-way splitter (14) for diverting the exhaust gas into one of three parallel legs (16), the exhaust gas flow in each leg being of equal velocity flow. Each leg (16) comprised a chamber (18) containing an oxidation catalyst (20) followed by a NOx-trap (22). The gas flows were then combined downstream of the NOx-traps and the total exhaust gas flow was passed through a "clean up" oxidation catalyst (24) to remove any unburned hydrocarbons (HC) exiting the NOx- traps before the exhaust gas was passed directly to the atmosphere. A fuel injector (26) comprising a fuel solenoid (28) was sited in front of each oxidation catalyst (20), a NOx sensor (29) in front of the exhaust splitter (14), combined NOx/air fuel ratio sensors (30) behind the NOx-traps and thermocouples (Tl, T2, T3, T4) measuring temperatures in front of and behind the oxidation catalysts (20) and at the exit of the reactors. The oxidation catalysts (20) and the NOx traps (22) were each coated on ceramic flow- through monoliths at 400 cells in"2 (62 cells cm"2) and 0.06 in (0.15mm) wall thickness. The oxidation catalysts (20) were 5.66 in (144mm) diameter x 3 in (76mm) and volume 75.5 in3 (1.24 litre), the NOx-traps (22) were the same diameter but 6 in (152mm) long and the "clean up" catalyst (24) 10.5 in (267mm) diameter x 3 in (76mm) long and volume 260 in3 (4.26 litres).

The experiments described here were conducted using one leg of the split exhaust only. The vehicle was operated using diesel fuel containing 50ppm sulphur and run at steady speeds of idle, 10, 20, 30 and 40 mph for periods of time; fuel was injected at each of these points and the air fuel ratio during injection determined as shown in Figure 4. The combination of time and duration (2 seconds injection, one per minute per leg) was selected empirically as it gave the best combination of exhaust gas temperatures (to maintain the NOx-trap within an active temperature window) and NOx conversion. Simultaneously the NOx emissions pre- and post- the system together with the temperature profiles were measured.

Figure 5 shows the NOx emissions (ppm) from the engine and after the NOx-trap for the idle condition together with the air fuel ratio measured after the NOx- trap. Figure 6 shows the temperature traces for the same period. From Figure 5 it is seen that when fuel is injected at the start of the idle period, the air fuel ratio drops from lean to rich as expected from the predictions in Figure 3 and, after the initial NOx breakthrough, good NOx conversion is seen. With time, the air fuel ratio remains lean throughout the injection event but good NOx conversion is still maintained. The exotherm (T2) generated over the oxidation catalyst helps maintain the temperature of the NOx-trap within its operating window of 220-550°C. An exotherm (T3) is also registered across the NOx-trap, some of which is caused by combustion of unreacted gaseous reductant from the oxidation catalyst. We interpret this result to mean that some of this exotherm is from the combustion of unburned fuel droplets reacting on the surface of the NOx-trap as time increases at this engine idle condition. This is because the system inlet temperature falls so as to be insufficient to vaporise the incoming fuel and the rear sensor measured air/fuel ratio spikes become less pronounced and more rounded, suggesting a sequence of the deposition, vaporisation and then the subsequent oxidation of the fuel droplets. The local richness caused by this event also serves to maintain the observed NOx-trap operation efficiency.

The results of the experiment with the bus held at a steady speed of 40mph are shown in Figures 7 and 8. Here the exhaust flow rate was much higher but the same injection flow rate was used as at idle and the exhaust was expected to remain lean throughout the injection periods (Figure 3). However, apart from the breakthrough spikes when the fuel is first injected, NOx is reduced over the remaining operating time, although not as efficiently as at idle. The exotherm (T3) over (T2) was sometimes lower than at idle, but because of the heat capacity of the increased flow rate of exhaust gases, it is very significant. Therefore an exothermic reaction is still taking place and again we believe that this is because some unburned fuel droplets are being carried through the oxidation catalyst and being combusted on the NOx-trap. The persistence of fuel droplets, despite the higher inlet temperature of the oxidation catalyst, is expected to occur because the greater exhaust flow rate that is likely to carry the droplets through the oxidation catalyst as is shown by the significant exotherm measured across the NOx-trap and the trap regeneration observed in apparently lean conditions.

Figure 9 presents the trend in calculated average NOx conversion efficiency, as a function of speed, for the system. Figure 3 indicates rich exhaust gas conditions do not occur above about 6 mph but good NOx conversions were obtained under lean conditions across a wider speed range. This is especially relevant in the range from idle to 30mph which is the most common operating range for an urban city bus.